Influence of supercell size on Gas-Surface Scattering: A case study of CO scattering from Au(1 1 1)

Abstract

We examine energy dissipation of impinging molecules (CO) on Au(1 1 1) using molecular dynamics on a machine learned high-dimensional potential energy surface (PES) describing both the molecular and surface degrees of freedom. The PES was trained using a neural network method from density functional theory energies and gradients obtained with a relatively small supercell size, but it is capable of providing an accurate description of the molecule-surface interaction for larger supercells. This property allowed us to investigate the dependence of the dissipation dynamics on the supercell size. Our simulations indicated that the supercell size has essentially no effect on the direct scattered molecules, but the energy dissipation of the trapping molecules is significantly influenced by the size of the simulation cell. This observation has important implications in understanding dissipation at the gas-surface interface and their effects on various dynamics processes.